How can you optimize native image build time for large applications?

Native image builds of large enterprise codebases can take many minutes, but several levers reliably cut that down.

1. Increase build parallelism and memory with -J-Xmx and letting the builder use more CPU cores - the points-to analysis phase is the most parallelizable part.
2. Reduce reachable surface area by trimming unused dependencies; every extra reachable class adds analysis and compilation work.
3. Use the reachability metadata repository instead of manually maintained broad reflection configs, since overly permissive "allDeclaredMethods": true entries pull in far more code than needed.
4. Enable Quick Build Mode (-Ob) during local development, which trades some runtime performance for a significantly faster build - reserving the fully optimized build for CI/release.
5. Use a build cache / layered images where supported, so unchanged dependency code doesn't get re-analyzed on every build.

In CI, splitting the reflection-agent tracing run from the final production build also avoids paying analysis costs twice.